Introduction
School-based speech-language pathologists address the development of oral and written language among students with communication disorders, including writing processes such as spelling (American Speech-Language-Hearing Association, ASHA, 2001, 2010). The ability to spell is one essential component of writing and written language competency (Berninger, Abbott, Nagy, & Carlisle, 2010; Devonshire & Fluck, 2010). Many children with language learning disabilities or expressive phonological impairments demonstrate difficulty with spelling (Clarke- Klein & Hodson, 1995; Dockrell, Lindsay, & Connelly, 2009; Dockrell, Lindsay, Connelly, & Mackie, 2007; Nathan et al., 2004; Silliman, Bahr, & Peters, 2006). Unfortunately, many speech-language pathologists do not perceive themselves as knowledgeable about writing and spelling, and therefore self-report limited confidence in working with children that have written language disorders (Blood, Mamett, Gordon, & Blood, 2010).
One of the factors contributing to the decreased confidence among speech-language pathologists may be limited experience applying what they know about oral language to written language, specifically in the area of spelling. This article describes a criterion-referenced approach to spelling assessment, the POMplexity for Roots and Affixes, based on the POMplexity (Benson-Goldberg, 2014) and modified POMplexity (E. Silliman, personal communication, November 28, 2015). The approach builds on skills speech-language
pathologists possess in the area of analyzing oral language samples and applies those skills in the area of spelling to evaluate, guide treatment, and support progress monitoring using the writing samples students produce every day in the classroom.
The Importance of Spelling
Spelling ability not only influences the correct spelling of individual words, but also other aspects of written literacy. For example, in a longitudinal study of typically developing writers, Abbott, Berninger and Fayol (2010) reported that spelling skills are related to and predictive of text composition among students in first through seventh grades. Poor spellers may need to devote more cognitive resources to spelling, which negatively influences other aspects of written composition (Hutcheon, Campbell, & Stewart, 2012; Puranik & Al Otaiba, 2012). When
students lack confidence in their spelling ability, they are more likely to select less appropriate, but familiar words (Kohnen, Nickels, & Castles, 2009), or disengage from writing all together (Lowe & Bormann, 2012). Brain imaging studies suggest that poor spellers may even
experience heightened stress when attempting to spell unfamiliar words (Richards, Berninger, & Fayol, 2009).
Traditional Views of Spelling
Spelling has traditionally been considered a rote mechanical skill (Bahr, Silliman, & Berninger, 2009). This perspective continues to dominate in educational settings, resulting in teachers using traditional memorization approaches to spelling instruction (Fresch, 2007). In stage models of spelling, student spelling sequentially progresses through distinct stages that are first influenced by phonological knowledge, then orthographic knowledge, and finally
spelling assessments influenced by these stage models only evaluate student responses as correct or incorrect (Masterson & Apel, 2013a; Treiman & Bourassa, 2000), but some further analyze errors and classify spelling errors according to the specific stage of development they reflect (Bear et al., 2012; Ganske, 2000).
More Recent Views of Spelling
More recently theorists have argued that spelling development is neither stage-based or linear, rather, it reflects the simultaneous integration of multiple sources of linguistic knowledge which deepen in development over time (Apel & Masterson, 2001; Devonshire & Fluck, 2010; Garcia, Abbott, & Berninger, 2010; Masterson & Apel, 2010a, 2010b; Rittle-Johnson & Siegler, 1999). Triple Word Form Theory proposes that across all stages of development, spellers attend to and coordinate orthographic, phonological, and morphological word forms (Richards et al., 2009). Supported by brain imaging (Berninger et al., 2010; Richards, Aylward, Berninger, et al., 2006) as well as behavioral studies (Berninger, Raskind, Richards, Abbott & Stock, 2008; Garcia et al., 2010; Nagy, Berninger, & Abbott, 2006), Triple Word Form Theory proposes that changes occur in the way phonological, orthographic and morphological word forms are involved over the course of spelling development, and that intervention can influence the coordinated
efficiency of these three components (Berninger et al., 2010; Richards et al., 2009). Linguistic Contributions to Spelling
While the three areas of phonology, orthography and morphology are integrated, and each is attended to in a coordinated way during spelling, difficulties with one or more of these linguistic areas contributes to difficulties with spelling. Phonological awareness, or the ability to segment and manipulate units at the sub-syllable level, is predictive of many literacy outcomes, including spelling (Bird, Bishop, & Freeman, 1995; Bradley & Bryant, 1983; Bryant, MacLean,
Bradley, & Crossland, 1990; Wood & Terrell, 1998). Phonological awareness is thought to reflect increased awareness of the internal structure of words, promoting successful phoneme to grapheme encoding required in spelling (Bourassa & Treiman, 2001). Therefore skills such as phoneme deletion (Muter & Snowling, 1997) and parsing words into phonemes (Ritchey & Speece, 2006) have been predictive of spelling outcomes, while phoneme manipulation
(Strattman & Hodson, 2005) and phoneme isolation (Caravolas, Hulme, & Snowling, 2001) have explained variance in spelling achievement.
Orthographic knowledge consists of the storage and representation of spoken language in its written form (Apel, 2011). Apel suggests this knowledge includes the stored mental
representations of specific grapheme sequences as well as generalized understandings of the rule- based ways orthographic patterns represent sounds. The mostly implicit awareness of rules for graphemes and grapheme combinations facilitates the use of statistical patterns in spelling, reducing or eliminating the need for memorization of individual words (Berninger, Abbott, et al., 2006; Deacon, Conrad, & Pacton, 2008). Students also draw from orthographic knowledge to produce plausible orthographic representations of unfamiliar words (Apel, 2011; Wolter & Apel, 2010). Orthographic knowledge has been found to contribute uniquely to spelling development (Walker & Hauerwas, 2006).
Morphological awareness includes the graphemic representation of morphemes, as well as the rules for affixing morphemes (Apel, 2014). Morphological relationships provide
additional statistically predictable patterns, and students increasingly use this knowledge to spell derived morphological words (Berninger, Raskind, Richards, Abbott & Stock, 2008; Carlisle,
accuracy (Deacon, Kirby, & Casselman-Bell, 2009; Walker & Hauerwas, 2006; Wolter, Wood, & D’zatko, 2009). Morphological awareness also assists the systematic storage of information into the mental lexicon (Nagy et al., 2006). Morphological awareness is highly correlated with spelling ability among elementary school children (Garcia et al., 2010) and uniquely explains variance in spelling outcomes among upper elementary and middle school students (Nagy et al., 2006).
Assessing Spelling Skills
Traditional standardized spelling assessments that only evaluate student responses as correct or incorrect do not provide speech-language pathologists with sufficient diagnostic information to support student improvement in spelling (Bear et al., 2012; Sharp, Sinatra, & Reynolds, 2008). Approaches that focus on stages of spelling development provide more information (Bear et al., 2012; Ganske, 2000), but presume strengths and weaknesses in linguistic foundations based on assigned stages. More recent multilinguistic approaches acknowledge the possibility of different profiles of underlying linguistic foundations among spellers, and support the prescriptive identification of specific deficits based on hypothesis- driven analyses of misspellings (Masterson & Apel, 2010a, 2013b; ).
Spelling assessments are most often administered by dictating words within a sentence context. Dictation spelling tests reliably capture a student’s spelling ability (Kohnen et al., 2009) while minimizing cognitive load in composition planning and execution. Unfortunately,
dictation spelling tests alone are unable to create a comprehensive profile of a student’s spelling ability (Hammond, 2004). In contrast, writing samples allow spelling achievement to be
evaluated in contexts that are more representative of real life literacy tasks that require planning, syntax creation, cohesion, vocabulary selection and monitoring.
A few spelling measures have been developed that can be used to assess linguistic errors of misspelled words within contextualized written compositions. The Spelling Sensitivity Score (SSS, Masterson & Apel, 2010b) divides words into elements which are defined as phonemes, juncture changes, and affixes, and then assigns a score based on the orthographic legality of the element. The authors suggest that the average element and word scores may indicate the phonologic or orthographic nature of underlying deficits, but this metric does not include separate examinations of phonological or morphological knowledge on spelling. In addition, some attempts to spell words are considered unanalyzable (e.g. two adjacent omissions) and cannot be included in analyses.
The Phonological, Orthographic, and Morphological Assessment of Spelling (POMAS, Silliman, Bahr, & Peters, 2006) is a qualitative scoring system that classifies errors into the three categories of phonology, orthography, and morphology and then further codes errors into one of 41 linguistic features associated with general American English. While the coding system documents the underlying linguistic features of each spelling attempt and how those linguistic features change over time, the approach is complex and time consuming. The POMplexity (Benson-Goldberg, 2014) and modified POMplexity (E. Silliman, personal communication, November 28, 2015) were developed as quantitative metrics to examine the misspellings of morphologically complex words, based on the qualitative descriptors identified in the POMAS. Examining spelling at the word level, the POMplexity and modified POMplexity separately rank the phonological, orthographic and morphological contributions based on the severity of the misspelling. Both of these frameworks are limited to analysis of morphologically complex
In contrast, the Multi-linguistic Coding System includes a linguistic analysis of errors in the domains of phonology, orthography, mental graphemic representation, morphology, and semantics (Bowers, McCarthy, Schwarz, Dostal & Wolbers, 2014). Words are the unit of analysis, and each misspelled word can be coded for one or more linguistic errors. One of the limitations of this assessment is that there are no references or guidelines provided for evaluating criteria such as pronunciation or legality, and error categories are vaguely defined.
There is clearly a need for an approach to analyzing spelling errors that is efficient enough for busy clinicians to use across all morphological word types, but with sufficient guidelines and specificity for supporting reliability. It is also important that the spelling
assessment sufficiently captures the full range of spelling errors students make when writing, in order to provide prescriptive information regarding the underlying linguistic contributions of phonology, orthography, and morphology.
POMplexity for Roots and Affixes: A New Approach to Spelling Assessment
The POMplexity for Roots and Affixes (Quick & Erickson, 2015) is a modification of the POMplexity (Benson-Goldberg, 2014) and modified POMplexity (E. Silliman, personal
communication, November 28, 2015) intended to address the need to efficiently analyze both monomorphemic and multimorphemic words. The POMplexity and modified POMplexity both require analysis of each phoneme and grapheme, but this new adapted approach evaluates errors at the level of the morpheme. By changing the unit of analysis to morphemes, and developing criteria that can be applied to both roots and affixes, the approach permits assessment of single morpheme, compound, contracted, inflected and derivational words. At the level of the
the need for more complex computations at phonemic or graphemic sublevels and making the approach more useful in clinical settings.
The POMplexity for Roots and Affixes separately analyzes each root or affix and then allows for the identification of categorical errors in the domains of phonology, orthography, and morphology. Appendix 1 provides a definition of each linguistic category and examples to accompany the criteria for phonological, orthographic and morphological errors with the
POMplexity for Roots and Affixes. For example, the misspelling of “burd” for “bird” would be coded as follows: (a) Phonology - plausible, (b) Orthography - legal vowel error, and (c) Morphology - non-morpheme substitution. Frequency counts for each categorical error within each linguistic area are calculated. The tool provides flexibility in analyzing patterns of spelling errors of roots and affixes. For example, frequency distributions can be calculated based on all morphemes, categories of morphemes (e.g., roots, affixes), or specific types of morphemes (e.g. inflectional morphemes). In order to control for differences in the number of misspelled words across sampling contexts, the total frequency count can be divided by the total number of errors, or by the total number of morphemes in the analysis, in order to produce a standardized metric for use in identifying changes in the relative contribution of each error type over time.
One of the weaknesses of previous studies is the vagueness of definition for phonological plausibility and orthographic legality. In POMplexity for Roots and Affixes, phonological plausibility and orthographic legality are determined according to the criteria outlined by Olson and Caramazza (2004). Specifically, phonological plausibility is determined by first identifying phoneme to grapheme(s) correspondences within the morpheme of a misspelled word. The
misspelling to be produced like the target. The phonological judgments are phoneme sensitive rather than context sensitive as outlined by Olson and Caramazza. Therefore “stoped” is considered phonologically plausible even though, contextually, doubling the consonant is required to maintain a short vowel in the medial position of a word.
In order to examine orthographic legality, words are first parsed into word-initial consonants or vowels (e.g., top, about), word-medial onsets (e.g., about), word medial-vowels (e.g., about), word-medial codas (e.g., harmony), and word-final consonants or vowels (e.g., about). When examples of orthographic letter or letter sequences in these categories found in the misspelled words can be identified among words in the Merriam-Webster Elementary
Dictionary, the response is considered legal (e.g., “rite” for “right”; “plant” for “planet”). Morphological plausibility is ascribed when the morphograph (i.e., written representation of morpheme) contains minor orthographic errors that suggest awareness of the target
morpheme. These errors include spacing errors (“some where” for “somewhere”) and
transpositions (“toegther” for “together”), as well as alternative orthographic representations of the target morpheme. Alternative orthographic representations include the range of spelling patterns for a target morpheme that occur in root, inflected and derived word forms. For example, the spelling patterns of “motive” and “motiv” are alternative orthographic
representations of the same target morpheme (motive; motivating). Therefore “motiveating” for “motivating” is considered morphologically plausible. Similarly, the spelling patterns of “stop” and “stopp” are alternative orthographic representations of the same target morpheme (stop; stopping), and therefore “stoping” for “stopping” is considered morphologically plausible. A score sheet for the entire process is provided in Appendix B.
Applying POMplexity for Roots and Affixes to Assessment
To demonstrate the utility of the POMplexity for Roots and Affixes, we randomly selected the writing samples of four 5th grade students who participated in a larger study of writing (Erickson, Geist, & Hatch, in press). The students completed the Story Construction subtest of the Test of Written Language 3 (Hammill & Larsen, 1996) as a pre- and post-test measure in the larger study. Students were instructed to write a story in response to a picture from a prehistoric period. The pre-test writing samples were transcribed and each word was coded as being either a single morpheme, compound (e.g., single unit of two words), inflected (free standing root or base with an affix that changes the tense, person, or number such as “girls” or “walked”) or derived (e.g., free standing root or base with one or more affixes that change the meaning or grammatical class of the base word such as “unhappy” or “growth”). Next
misspelled words for each student were analyzed for phonological, orthographic, and
morphological errors using the POMplexity for Roots and Affixes in order to provide specific examples of the application of the approach. The scoring of misspelled words for each student is provided in Appendix C.
Participants
Participants from the larger study were selected for the current study on the basis of having writing samples that appeared to represent diversity in the number and types of spelling errors. The selected participants included four males from two separate fifth grade classrooms ranging in age from 10 to 11 years. They represented diverse backgrounds, and half were eligible for free price or reduced lunch. One student had a diagnosis of high functioning autism,
Table 2.1 Participant Demographics
Student Chronological Age Ethnicity Free/Reduced Lunch Disability
1 10;3 Hispanic Yes -
2 10;3 white No -
3 11;2 black Yes -
4 10;0 white No Autism
The student writing samples varied in length and in the number of spelling errors. Students 1 and 2 wrote compositions of similar length and had 10 or fewer errors. Students 3 and 4 wrote the longest and shortest compositions, respectively, yet demonstrated a similar total number of errors (n = 25-28). Across all four students, the compositions had a mean of 93 total words in length, with an average of 24% of words being misspelled. The writing sample of student 1 had a high percentage of single morpheme words with the majority of misspellings occurring among single morpheme words. Student 2 used the greatest number of
morphologically complex words, and had no errors with single morpheme words. Student 3’s writing sample contained no derivationally complex words and his errors were distributed across single morpheme, compound and inflected words. Student 4 had the greatest percentage of misspelled words (60%), and all words that were not single morphemes were misspelled. The total number of words and errors for each of the students are reported below in Table 2.2. Monomorphemic Level of Analysis
Using the POMplexity for Roots and Affixes, the first analysis examined the distribution of categorical errors among the roots of misspelled words that were monomorphemic. In
addition to single morpheme words, this grouping included the parsed morphemes of compound and contracted words. Compound words and contracted words were included in the
Table 2.2 Total Number of Words and Total Number of Misspelled Words across Morphological Word Types
Distribution of Errors
Total Words Derived Inflected Compound Single
Student 1 94 (10) 1(0) 9 (2) 2 (2) 82 (7)
Student 2 92 (5) 3 (2) 14 (2) 5 (1) 70 (0)
Student 3 141 (25) - 14 (9) 9 (4) 118 (12)
Student 4 47 (28) - 5 (5) 1 (1) 41 (22)
monomorphemic analysis as they are composed of two single morpheme words that form a single word unit. Compound words do not include changes in pronunciation or spelling when combined (e.g. nobody, somewhere), but contracted words include an apostrophe that marks the changes in pronunciation and spelling of the second morpheme (e.g., can’t, don’t).
Among monomorphemic roots, student 2 demonstrated little difficulty as he only had one misspelled morpheme. The remaining students had 11 to 25 misspelled monomorphemic roots. When examining phonology, all of student 1’s misspellings were either phonologically correct or plausible. In contrast, students 3 and 4 made one or more phonological errors in approximately 33-50% of the monomorphemic words they wrote. Student 3 tended to use substitutions or omissions while student 4 tended to use substitutions or additions. In the area of orthography, the errors of student 1 were fairly evenly divided between minor errors of spacing or
transpositions, and legal grapheme errors of vowels. The orthographic scoring of student 3’s monomorphemic words was fairly evenly distributed across correct, minor, legal and illegal
of his errors involved both vowels and consonants. In the area of morphology, nearly one half of the errors student 1 made were morphologically plausible as they primarily involved spacing errors among compound words. The errors student 3 made reflected awareness of
morphographs, as there were frequent substitutions of homophones, other morphemes, and alternative orthographic representations of the target morpheme. The errors student 4 made primarily included non-morpheme substitutions, indicating more limited morphographic awareness. The frequency counts and percentages of categorical errors for monomorphemic roots are reported in Table 2.3.
Interpreting Monomorphemic Scores. The variations in scores across the domains of phonology, orthography, and morphology suggest that the three students who struggled with spelling monomorphemic words would benefit from spelling instruction that focused on strengthening different underlying linguistic skills. For example, most of the errors student 1 made involved either phonologically plausible misspellings of polysyllabic words or spacing errors with compound words. Therefore student 1 would benefit from intervention that increased his orthographic awareness among these particular word types. The multiple phonological errors made by students 3 and 4 indicate a need for increased phonological awareness. In addition, student 3 would benefit from morpheme-based instruction that assists him in differentiating between morphographs that have similar phonological or orthographic skeletons, while student 4 requires intervention that assists him in the spelling of morphographs using legal orthographic